The pragmatics of parenthetical constructions : evidence from English and Polish

19  Download (0)

Full text


Jagiellonian University in Kraków


1. Introduction

In this article, I address some problems that parentheticals, a group of formally dissimilar but functionally uniform linguistic expressions, pose for a seman- tic/pragmatic analysis. On the basis of the empirical data drawn from English and Polish (for comparison and to verify the theoretical claims), I re-examine the semantic/pragmatic properties of parentheticals and approach them from a relevance-theoretic perspective that introduces the distinction between con- ceptual and procedural meaning that linguistic expressions, parentheticals included, may potentially encode. I also investigate whether a relevance-the- oretic account of parentheticals may help in a better understanding of paren- theticals as a class of linguistic expressions and how it may clarify their status in linguistic analysis and theory.

2. Parentheticals in linguistic research: the state of the art

Linguistic communication is peppered with parentheticals. Whether in speech or writing, utterances are frequently interrupted by more or less complex par- enthetical expressions, typically occurring in the middle of a linguistic struc- ture (referred to as its host), “unintegrated in the sense that [they] could be omitted without aff ecting the rest of that structure or its meaning” (Biber et al. 1999: 1067), marked by punctuation (commas, parentheses/brackets, or dashes) in texts, and set off by comma intonation in speech.


With the defi nition formulated as above, parenthetical functions as an um- brella term for a wide spectrum of linguistic expressions and may include, among others, the following:1

(1) Th e driver of Al-Kindi’s only remaining ambulance – the other three had been stolen or looted – has disappeared. So the dangerously ill Mr Khas- sem was bundled into a clapped-out rust-bitten Moskvich 408 (Blakemore 2006: 1671).

(2) A helicopter, a helicopter – and there was me who’d never even fl own in an ordinary plane – would come and pick me up at … (Blakemore 2009: 11) (3) What is obvious – and we have eye-witness reports – is that they were killed

(Blakemore 2009: 11).

(4) Margaret Wynne Nevinson, an active feminist, who was a member of a school management committee for twenty-fi ve years and who also served as a Poor Law Guardian, found her fellow male Guardians actively hostile.2 (5) Margaret Wynne Nevinson, an active feminist, who was a member of

a school management committee for twenty-fi ve years and who also served as a Poor Law Guardian, found her fellow male Guardians actively hostile.3 (6) In fact, if you really want to know, I think of you as a very attractive man.

(7) You’ll need lots of patience, but as you say your boy has a good nature oth- erwise, the chances are you’ll succeed.

(8) It is, aft er all, what we brought you here for – to resolve our problems for us.

(9) Th e colleges of South Kensington were involved in this along with other groups which, unfortunately, are passed over in this account.

Th ough intuitively and relatively easily recognized in utterances, parenthet- icals have proved to be a challenge for linguistic theory, where they have been analyzed from various perspectives: syntactic, semantic/pragmatic, or prosod- ic. As a result the studies available are hardly comparable and usually focus on diff erent sets of phenomena. Th ere is neither agreement in the literature on the criteria determining parenthetical status, nor consistent terminology used in reference to parenthesis.

Semantic/pragmatic studies are no exception, as there is no consensus in semantic/pragmatic theory on how the contribution of parenthetical expres-

1 Parentheticals are given in italics and represent both written and spoken sources.

2 Th e English examples provided here are from the British National Corpus (http://www., unless specifi ed otherwise.

3 In fact, it seems justifi ed to claim that the entire: ‘an active feminist, who was a member of a school management committee for twenty-fi ve years and who also served as a Poor Law Guardian’ can be considered parenthetical. More than one parenthetical can be associated with a single sentential structure (Espinal 1991).


sions to utterance interpretation and their discourse function should be best approached.

In the traditional speech-act accounts, it has been generally assumed that parenthetical expressions have no propositional meaning, i.e., they do not in- teract semantically with the utterance they are embedded in and do not aff ect the proposition expressed by the host. Rather, they indicate its illocutionary force (Urmson 1952), and comment on the main proposition. As stage-direc- tions of some sort, parentheticals function as signals to guide the hearer to a proper appreciation of the host.

Th e traditional speech-act approach to parenthetical insertions stands in contrast to the perspectives taken by linguists analyzing language as a form of social interaction, where parentheticals are taken to be examples of disfl uency characterizing unplanned discourse. In such approaches, as Wichmann (2001:

189) puts it, parentheticals are nothing but:

hesitations, revisions and self corrections, incidental comments about what is be- ing said in the host utterance, self-addressed questions and reminders, responses to something external to the conversation, and questions designed to elicit feedback or to check attention and as such, they are evidence that speakers have trouble planning their utterances, but are constrained by interactional principle to keep talking.

However, the discourse/pragmatic integration of the parenthetical with the host utterance or the lack thereof does not seem to be the only concern for parenthetical description in semantic/pragmatic studies. Given the existence of pragmatic links between the parenthetical and its host, it seems desirable to specify the range of their pragmatic functions. Th e problem is that, as already suggested, like other pragmatic markers, parentheticals can have any number of functions depending on the context and moreover, these functions may sometimes overlap. In consequence, it is not easy to reduce parentheticals to a single pragmatic meaning (Rouchota 1996, 1998). Bolinger (1989: 190), for instance, lists three main relationships that a parenthetical can have with refer- ence to its host, which he classifi es as:

• comments (oft en providing additional information or aft erthought), e.g., I think, I mean, I believe;

• revisions, e.g., or, that is, rather;

• decisions e.g., like, well, let’s say.

Th us, although it is traditionally assumed that there is a class of expres- sions that can be grouped together as parentheticals, intuition is oft en relied on when deciding what to include in the parenthetical class. Indeed, classifi cation is not an easy task, taking into consideration the fact that when a parenthetical is taken independently of its host utterance, there is no single criterion which can be chosen to identify it.


In view of this, pragmatic research on parentheticals has two main goals:

fi rstly to explain the use of parentheticals in actual discourse where paren- thetical insertion seems to be to be a performance phenomenon, constituting disfl uency typical of unplanned discourse and marked by special prosody, as in the examples in (1)‒(3), and secondly to account for a class of parenthetical constructions intervening within a clause which are realized with the same comma intonation, presumably hold similar discourse functions, but which are taken to be licensed by grammar, as in the examples in (4)‒(9).

Last but not least, since present-day linguistic research is characterized by broadening of the fi eld to include new phenomena and there is more interest in constructions which are less prototypical, the defi nition of a parenthetical in terms of lack of syntactic, semantic and prosodic integration is not satisfactory. It has been shown that certain syntactic relations between the parenthetical and its host do exist: e.g., some binding eff ects and scope relations can be observed be- tween the parenthetical and the host (Dehé and Kavalova 2007: 4). Furthermore, it has been argued, most notably by Blakemore (2006, 2009), that parentheti- cals do have propositional meaning and may contribute to the interpretation of the utterance they are embedded in by altering the context for its interpreta- tion, which puts the non-truth conditionality of parentheticals in question. Fi- nally, any of the important prosodic features can be suspended depending on the function of the parenthetical, its length and position (Bolinger 1989; Dehé and Wichmann 2010; Kaltenböck 2008, a.o.). Th is suggests that none of the features traditionally linked with parenthesis qualifi es as a necessary condition.

Th us, the pragmatic function(s) that parentheticals can have with respect to their host utterances emerge(s) as the only one positive and reliable feature characterizing this class of linguistic expressions. Th erefore the semantic/prag- matic relationship between the parenthetical and the host is worth re-inves- tigating with a view to fi nding the ground for a unifi ed pragmatic account of parenthetical phenomena.

3. Parentheticals in semantics and pragmatics: (non-)truth conditional meaning vs. non-unitary theory of semantics

As already mentioned, in the traditional speech-act accounts of parenthetical phenomena, their semantics has been explained in terms of non-truth condi- tional meaning. In this section, I focus on the meaning of parentheticals and the way they might contribute to utterance interpretation from a relevance- theoretic perspective, and more specifi cally, a non-unitary theory of semantics proposed by Blakemore (1987, 2006, 2009). Th is approach is discussed with respect to empirical data from English and Polish to see whether it has any


advantages over the traditional speech-act account and whether it might help clarify the semantic/pragmatic properties of parentheticals.

3.1. Non-truth conditional meaning

As observed by Grice (1989: 362), in the sentence such as (10) below, the meaning of the phrase ‘on the other hand’ diff ers from the meaning of phrases such as ‘brother-in-law’ or ‘great aunt.’ ‘On the other hand’ does not have a rep- resentational meaning, i.e., the meaning that encodes a concept and is part of the proposition expressed in the utterance, unlike ‘brother-in-law’ and ‘great aunt.’ Rather, it shows that the proposition should be taken as a comment, in this case of the contrastive type, on the propositional content of the utterance, thus constraining its context of interpretation:

My brother-in-law lives on a peak in Darien; his great aunt, on the other hand, was a nurse in World War I (Grice 1989: 362).

Similarly to the contrastive ‘on the other hand,’ the parenthetical expres- sions ‘I think’ and ‘you know’ in (11) represent what Grice referred to as high- er-level speech-acts in the function of comments (showing how the speaker is committed to the proposition expressed in the utterance) on lower-order speech-acts:

(11) a. Th at was a nice picture – Marie liked it, I think, cos she didn’t mind me sticking it on the wall.

b. He is one of those players you want on your side because you know he will be superb week-in, week-out.

However, as Ifantidou (1994, 2001) convincingly argues, being parentheti- cal is not a necessary condition for being non-truth conditional. In the exam- ples below, the inserted expressions fi t the defi nition of a parenthetical given above. Yet the parenthetical use of ‘you say’ in (12) contributes to the truth conditions of the proposition expressed in the utterance since the speaker will be understood to be committed to (13a) but not to (13b):

(12) Because John is, you say, a spy, we should be careful what we say to him (Ifantidou 2001: 149).

(13) a. We should be careful what we say to John because you say he is a spy.

b. We should be careful what we say to John because he is a spy.

Th e examples shown above can be further compared to the ones in (14), which present the contrast fi rst observed by Quirk et al. (1985): (14b), unlike (14a), does not have the same truth conditions as the sentence ‘George is a liar’:


(14) a. George is, as you said, a liar (*but I don’t believe it).4 b. George is, you say, a liar (but I don’t believe it).

Th e solution to the problem of the non-truth conditionality of parentheti- cals off ered by Ifantidou (2001) builds on a hypothesis off ered to this prob- lem by Wilson and Sperber (1993). Simply put, the use of ‘you say’ marks the proposition expressed by the host clause as a proposition which is relevant as a representation of a thought which is not the speaker’s, i.e., it is an example of interpretive use (cf. also Sperber and Wilson 1995). As such, it will not be understood as the expression of the speaker’s commitment to the proposition but rather as indicating that the proposition is a faithful representation of an- other thought, i.e., the thought of the hearer. According to Sperber and Wilson (1995), the relationship between the proposition expressed and its interpreta- tion is not that of identity but resemblance of content. In fact, the speaker in (14b) might be expressing any of the following propositions:

(15) a. George never tells the truth.

b. Whatever George has told you will be a lie.

c. You should never believe what George says.

Th us, if the speaker using ‘you say’ as an example of interpretive use does not express his/her own commitment to the proposition, the inserted material will aff ect the truth conditions of the host. In (14a), on the other hand, the use of ‘as you said’ will be understood as the representation of the speaker’s thoughts. Th is suggests that ‘as you said’ is not truth-conditional in this case.

Ifantidou (2001) draws a distinction between parentheticals which do not af- fect truth-conditional content and those that do by marking it as a case of interpretive use or by aff ecting the strength of the proposition expressed. She does so with the help of two concepts: procedural and conceptual meaning, fi rst introduced into the theory of semantics by Blakemore (1987).

3.2. Conceptual vs. procedural meaning

According to both Blakemore (1987) and Wilson and Sperber (1993), elements with conceptual meaning are said to contribute to the content of assertions and are analyzed as encoding elements of conceptual representations. Procedural elements, on the other hand, encode information about how these representa- tions are to be used in inference, how these representations should be taken. In other words, while the former contribute directly to the construction of concep- tual representations, i.e., concepts, the latter indicate how to manipulate these representations, i.e., procedures in communication, which on the RT approach,

4 Th e denial in (14a) is inconsistent, as the ‘as’-clause implicates the truth of its content.


is an inferential process. What is of vital importance, however, is the question whether the recognition of two types of meaning may help in a better under- standing and description of parenthetical phenomena. What is more, it might be worth exploring if these generalizations carry over to other languages.

4. Discourse parentheticals and their pragmatic interpretation

Th e parentheticals in the examples given in (1)‒(3) have never been analyzed as grammatically integrated into the host structure. However, following Blake- more (2006: 1684‒85), the relationship between the parenthetical and its host in such examples can be captured by claiming that the parentheticals achieve relevance by contributing their own cognitive eff ects to the utterance interpre- tation, but in the context of assumptions made accessible by the interpretation of the host.

For example, in the English example (1) repeated below as (16) and in the Polish example in (17), the parenthetical and the host are related at the level of pragmatic interpretation. Each is interpreted as having its own relevance and the parenthetical does not contribute to a combined proposition whose relevance is greater than that of the parenthetical and host taken individually.

However, as Blakemore (2009: 11) argues, the parenthetical is not completely sealed off from the truth conditional content of the host. Its relevance rests on the information about the truth conditional content of its host. Th us, the par- enthetical identifi es the conceptual content of the host but it does not aff ect its implicit or explicit content:

(16) Th e driver of Al Kindi’s only remaining ambulance – the other three had been stolen or looted – had disappeared. So the dangerously ill Mr Khas- sem was bundled into a clapped-out rust bitten Moskvich 408 (Blake- more 2006: 1671).

(17) W prezydium Komisji Kultury zasiadał wówczas pan poseł Wełnicki – nie pamiętam imienia – sympatyczny skądinąd człowiek, który swoją rolę w tej komisji sprowadzał właściwie do odbijania telewizji publicz- nej, jak sądził, z jednego ugrupowania do AWS i tym się tylko w czasie obrad tej komisji – albo przede wszystkim tym się zajmował.5

‘Th e Culture Committee was chaired by MP Wełnicki – I don’t remem- ber his fi rst name – otherwise a nice man, who actually limited his

5 Th e Polish examples provided here are from the Narodowy Korpus Języka Polskiego IPI PAN (, unless specifi ed otherwise.


actions to retaking national television, as he thought, from one political party to the AWS party and it was the only thing – or the main thing he was concerned with.’

Th e parentheticals are pragmatically integrated with the host in the sense that they provide the answers to the question raised by the hosts (‘Why was there only one ambulance?’ in the English example and: ‘Why is the person, the head of the national culture committee, referred to only by his surname?’

in the Polish example). Th e relevance of the coordinated proposition is not greater than each of the conjuncts taken individually. Th is fact is supported by the impossibility of using ‘and’ to communicate pragmatic integration (cf.

Blakemore and Carston 2005):

(18) ?? Th e driver of Al Kindi’s only remaining ambulance – and the other three had been stolen or looted – had disappeared. So the dangerously ill Mr Khassem was bundled into a clapped-out rust bitten Moskavich 408 (Blakemore 2009: 12).

(19) ?? W prezydium Komisji Kultury zasiadał wówczas pan poseł Wełnicki – i nie pamiętam imienia – sympatyczny skądinąd człowiek, który swoją rolę w tej komisji sprowadzał właściwie do odbijania telewizji publicz- nej, jak sądził, z jednego ugrupowania do AWS i tym się tylko w czasie obrad tej komisji – albo przede wszystkim tym się zajmował.

‘Th e Culture Committee was chaired by MP Wełnicki – and I don’t re- member his fi rst name – a nice man by the way, who actually limited his actions to retaking national television, as he thought, from one political party to the AWS party and it was the only thing – or the main thing he was concerned with.’

A diff erent type of pragmatic integration of the parenthetical and its host is shown in Blakemore’s (2009: 11) example given in (20) and the Polish example given in (21):

(20) A helicopter, a helicopter – and there was me who’d never even fl own in an ordinary plane – would come and pick me up at …

(21) W parlamencie podniósł się wielki krzyk ‒ nawet konserwatywnie na- stawione panie poczuły się urażone, jakby ktoś nazwał je towarem z dru- giej ręki – więc ostatecznie projektu nie uchwalono.

‘An outcry arouse in Parliament – even pro-conservative women MPs felt off ended, as if somebody called them second-hand stuff – so eventu- ally, the proposed bill was not passed.’

In contrast to the previous examples, it seems that the parentheticals in (20) and (21) aff ect the interpretation of the host at the level of implicit communica-


tion. In particular, it might be proposed that the parenthetical has no relevance beyond its eff ect on the interpretation of its host in that it modifi es the context for the recovery of the implicit content of the host and contributes, together with the host, to the recovery of a single proposition, greater than either of them in- dividually. Th e parenthetical in the English example encourages the hearer to re-consider the contextual assumptions made accessible by the concept of the helicopter for the derivation of implicatures which capture the excitement of travelling in a helicopter for someone who has never fl own in any kind of plane at all (Blakemore 2009: 12). With respect to the Polish example, it can be argued that the hearer is expected to revisit the contextual assumptions made accessible by the concept wielki krzyk (‘an outcry’) and derive the implicature of MPs’ total criticism, as a result of which the proposed bill was not passed.

To see how parentheticals can aff ect the interpretation of their hosts at the explicit level, it is necessary to explain the diff erence between explicatures and higher-level explicatures introduced into the Relevance Th eory by Sperber and Wilson (1995).

In RT, explicatures are explicitly communicated assumptions, i.e., linguisti- cally encoded logical forms enriched in such a way as to express determinate propositions, as illustrated in (22a):

(22) a. Johni is at home.

b. Ann is saying that Johni is at home.

c. Ann is asserting that Johni is at home.

d. Ann thinks that Johni is at home.

Optionally, the proposition may be embedded under a higher-level descrip- tion: a speech-act description (22b) or a propositional attitude description ((22c) and (22d)). As such, they are referred to by Wilson and Sperber (1993) as higher-level explicatures. Higher-level explicatures, like logical forms and fully propositional forms are conceptual representations recovered by a com- bination of decoding and inference (Wilson and Sperber 1993: 11). While to obtain (22a) the hearer must decode the semantic representation of the utter- ance, to obtain the higher-level explicatures in (22b)‒(22d) the hearer must make additional inferences about the speaker’s attitude to the proposition that is being expressed and the type of speech-act that is being performed. Th is is because both explicatures and higher-level explicatures have their own truth conditions and, therefore, are capable of being true in their own right. Only the proposition expressed, however, contributes to the truth conditions of the associated utterance. Th e higher-level explicature will not be part of the truth- conditional content of the host utterance.

In view of this, the parentheticals illustrated in (23) and (24) might be said to aff ect the interpretation of the host at the level of explicit content. Th ey


might be said to contribute to higher-level explicatures; in other words, their relevance is in the eff ect they have on the hearer’s understanding of the degree of commitment being communicated by the host.

(23) What is obvious – and we have eye-witness reports – is that they were killed (Blakemore 2009: 11).

(24) Tak się akurat składa, że owe wszystkie informacje czerpię z tych samych, lub jak przypuszczam, chyba tych samych, źródeł, jak mój adwersarz.

‘It seems that I obtain all the information from the same, or at least I’m assuming that they are the same sources, as my opponent does.’

In (23), the hearer is intended to recover a higher-level explicature con- veying a greater degree of commitment to the proposition expressed than any higher-level explicature possible to be recovered otherwise, in (24), on the other hand, the hearer is intended to recover a higher-level explicature which conveys less certainty towards the truth of the proposition expressed (Blake- more 2009: 13‒14).

In (25) shown below, the parenthetical contributes explicitly to the inter- pretation of the host as well, but this time, by specifying how the quantifi er

‘każdy’ (‘every’) and its reference should be identifi ed. Th us, it enables the hearer to make hypotheses about the relevance of the information communi- cated by the host. In particular, it appears that the parenthetical specifi es how its domain should be interpreted.

(25) Każdy przedsiębiorca spełniający warunki, to znaczy ten, kto poniesie nakłady inwestycyjne nie mniejsze niż 100 000 EUR, będzie mógł tam zainwestować.

‘Every businessman who will meet the demands, that is the one who will invest no less than 100,000 EUR, will be allowed to invest there.’

Further, the parenthetical in (26) may achieve its relevance at the level of explicit content of the host by taking part in what may be labeled as on-line concept construction (Carston 2002). To begin with, the hearer decodes the meaning of ‘przedtem’ (‘before’). Th e parenthetical, on the other hand, will communicate the information that will encourage the hearer to use contextual assumptions further to recover the concept the speaker wishes to communi- cate by uttering ‘przedtem’ as the period of three years they lived in Ostrowiec aft er leaving Warsaw:

(26) Przedtem – to znaczy w ciągu trzech lat, które przeżyli w Ostrowcu po opuszczeniu Warszawy – mieszkali w śródmieściu, w reprezenta- cyjnej Alei Trzeciego Maja i w bliskim sąsiedztwie koszar, w których stacjonował głośny pułk ułanów ostrowieckich.


‘Before – that is during the three years they spent in Ostrowiec aft er leaving Warsaw – they lived in the city centre, in the prestigious Aleja Trzeciego Maja and in the close vicinity of the barracks, in which a well- -known uhlan regiment of Ostrowiec was deployed.’

To sum up, it seems that, contrary to the existing accounts, the discussed examples should not be treated as instances of disfl uency or explained in terms of interactional principles which require speakers to keep talking (Wichmann 2001: 189). Rather, as Blakemore (2006, 2009) argues, within a relevance-the- oretic account, they can be explained in terms of the pursuit of relevance. Fol- lowing Sperber and Wilson (1995: 204) it might be observed that the fact that an utterance is produced and processed over time means that a hearer will be able to access some of its constituent concepts, with their associated logical and encyclopedic entries, before others. Th is means that certain contextual as- sumptions will be triggered before others, and that the hearer, who is assuming optimal relevance, will use these to construct hypotheses about the speaker’s informative intention. In all of the examples discussed above, the use of the disrupted structure is consistent with the speaker’s aim of achieving relevance for a minimum cost in processing eff ort.

Th is might lead to the preliminary observation that true parentheticals will always carry some relevance to the interpretation of the host. It should not be assumed, however, that discourse parentheticals are truth-conditional. At the same time, though, they are not completely sealed off from the truth condi- tions of their hosts. If truth conditions can be roughly understood as “the truth conditions of the thought communicated by the host” (Blakemore 2009: 16) and this thought is to be recovered from the encoded semantic representation by pragmatic inference, it seems that the parenthetical aff ects truth conditions at the level of pragmatic interpretation. Th e relevance of parentheticals lies in the role they play in the pragmatic enrichment of their hosts and they have no relevance beyond this role.

5. Parenthetical comment clauses: concepts or procedures?

5.1. Parenthetical comment clauses: a challenge for syntactic and semantic/pragmatic analyses

Having tried to clarify the exact role of pragmatics in the interpretation of discourse parentheticals, it will be interesting to see whether the pragmatic interpretation of the parentheticals introduced in (4)‒(9) can be captured in


similar relevance-theoretic terms. Th ese parentheticals diff er from the ones in (1)‒(3) in that they have been treated as grammatical phenomena since, fol- lowing Espinal’s (1991: 727) observation that

it seems obvious that our knowledge of a natural language also tells us whether a given substring that apparently occurs within a syntactic unit is syntactically in- dependent of the rest of the string or whether it is incorporated into a single phrase marker, and that this knowledge should be accounted for by linguistic theory.

Th e intuition that the grammar generates linguistic structures with em- bedded parenthetical constituents is shared by many linguists, among others, by Taglicht (1996: 195), who claims that grammatical parenthetical phenom- ena must be distinguished from diversions and intrusions which characterize spontaneous discourse and there must be something that justifi es generating an utterance with a parenthetical but not the utterance of ‘Come in’ in the middle of the sentence when hearing a knock on the door. Th is assumption has implications for analyzing parentheticals in linguistic theory. In particu- lar, accommodating syntactically independent, though linearly ordered par- entheticals in syntax, in which the notions of precedence and linear order play a vital role, has been a considerable challenge. Further, due to their status as a syntax/pragmatics phenomenon, it is still debatable whether their proper- ties stem from their grammar or whether they have a purely pragmatic expla- nation.

In the middle of this debate come parenthetical comment clauses that seem to be central in the description of parentheticals due their syntactic and se- mantic properties. On the one hand, they are clausal in nature and are taken to contain representational meaning, just as discourse parentheticals do; on the other, they are argued (e.g., Brinton 2008) to be non-truth conditional, phonologically short, procedural pragmatic markers, similar in behavior and function to discourse markers. Th us, they seem to be suitable candidates for being reinterpreted in terms of conceptual or procedural meaning within a rel- evance-theoretic framework.

5.2. Comment clauses in speech-act accounts

Since the infl uential Urmson’s paper (1952), the verbs ‘think,’ ‘know,’ ‘believe,’

‘mean,’ ‘suppose,’ etc. are referred to as parenthetical verbs and treated in the traditional speech-act literature as similar to illocutionary or attitudinal ad- verbs, i.e., indicating the speaker’s degree of commitment to the proposition expressed. Illocutionary and attitudinal adverbials do not contribute to the truth conditions of the utterances they are embedded in by all standards for non-truth conditionality.


Similarly, parenthetical comment clauses are argued not to contribute to the proposition expressed and since their only function is that of a formula or an indicator (Austin 1962) of the performance of the act, they fi t the speech- act semantics just as other illocutionary force indicators, the above-mentioned adverbials or, say, performative verbs.

Th us, on the standard speech-act account, the parenthetical comment clauses in (27b)‒(27d) would be considered stylistic variants (‘I think’ is said to be loosely attached to the sentence it accompanies and can be reordered) that weaken the strength of the assertion expressed in (27a) and give rise to complex speech-acts:

(27) a. John is insane. [stronger]

b. I think John is insane. [weaker]

c. John is, I think, insane. [weaker]

d. John is insane, I think. [weaker]

However, as Ifantidou (1994: 197) observes, if the examples in (27b)‒(27d) involve complex speech-acts, i.e., two utterances (the one with the assertion and with the comment clause), there is no convincing explanation why each one could not have its own truth conditions and the parenthetical comment clause should be devoid completely of any descriptive content.

Further, the equivalent examples from Polish in (28) do not support the claim that all the sentences in (27) are stylistic variants. Crucial syntactic dif- ferences can be immediately noticed in (28a) if compared to (28b)‒(28c):

(28) a. Sądzę, że Jan jest szalony.

b. Jan jest, jak sądzę, szalony.

c. Jan jest szalony, jak sądzę.

Th e fi rst sentence cannot be taken to contain a true, syntactically uninte- grated parenthetical, since its syntactic representation would be that of a com- plex sentence with a transitive verb in the main clause followed by an object which is a subordinate clause introduced by the complementizer ‘that.’

Moreover, ‘I think’ in (27b) is truth-conditional. Th is can be supported by applying a truth-conditionality test, i.e., embedding the sentence with the ex- pression to be tested into a conditional and seeing if this expression falls within the scope of ‘if ’ (Ifantidou 1994: 198):

(29) If I think John is insane, he will not be arraigned or tried.

Th e question is, as Ifantidou (1994: 199) formulates it, under what cir- cumstances the speaker is claiming that John will not be arraigned or tried. If

‘I think’ makes no contribution to truth conditions, it should be synonymous with:


(30) If John is insane, he will not be arraigned or tried.

However, the two sentences are not synonymous. ‘I think’ does fall within the scope of ‘if ’ and is truth-conditional.

True parentheticals (i.e., stylistic variants, with the parenthetical comment clause in the sentence-initial, mid-sentence and sentence-fi nal position) are il- lustrated by the English examples in (31) and their Polish counterparts in (32), respectively:

(31) a. I think, John is insane.

b. John is, I think, insane.

c. John is insane, I think.

(32) a. Jak sądzę, Jan jest szalony.

b. Jan jest, jak sądzę, szalony.

c. Jan jest szalony, jak sądzę.

Th e truth-conditional tests confi rm this claim (irrespective of the position of jak sądzę ‘I think’ within the sentence). In the example (33) below, jak sądzę

‘I think’ does not fall under the scope of ‘if ’ and, in consequence, proves to be non-truth conditional:

(33) Jeśli Jan jest, jak sądzę, szalony, nie będzie oskarżony ani sądzony. (= (34)) (34) Jeśli Jan jest szalony, nie będzie oskarżony ani sądzony.

Moreover, parenthetical comment clauses seem to take the whole utterance in their scope (both in the English and the Polish examples below):

(35) If John is insane, he will not be arraigned or tried, I think.

(36) Jeśli Jan jest szalony, nie będzie oskarżony ani sądzony, jak sądzę.

5.3. Parenthetical comment clauses in a relevance-theoretic account: how do comment clauses comment?

Given the distinction between conceptual and procedural meaning that ex- pressions in a natural language may encode, the question that arises is whether parenthetical comment clauses fall on the conceptual side of this distinction, i.e., whether they encode constituents of conceptual representations and are similar to discourse parentheticals in this respect or whether they are proce- dural, i.e., whether they show how the associated utterances should be infer- entially processed.

As argued by Rouchota (1998), expressions that encode procedural mean- ing can be expected to exhibit properties indicative of their non-conceptual


status. Th eir meaning should be relatively diffi cult to bring to conscious aware- ness; such expressions also should not combine with other expressions to form expressions of greater semantic complexity in the way conceptual expressions do. Procedural expressions also should not be subject to relations of entailment or contradiction. Conceptual expressions, by contrast, have phonetic and syn- tactic representations and concepts as their constituents. Besides, they can en- ter into logical relations such as contradiction or entailment; they can describe or partially characterize a certain state of aff airs, they can be true or false, and they can act as input to inference rules.

Th e speaker using a parenthetical can be accused of making an untruthful claim (Ifantidou 1994: 202):

(37) A: John is waiting at the airport, I think.

B: Th at’s not true; you don’t think anything of the sort.

(38) A: Krytyk literacki nie mógłby, jak sądzę, zasnąć po lekturze tych wierszy.

‘A literary critic wouldn’t be able, I think, to fall asleep aft er reading these poems.’

B: Nieprawda, nie sądzisz tak wcale.

‘Th at’s not true, you don’t think so.’

Examples like the above thus suggest that although comment parentheti- cals do not contribute to the truth conditions of the utterance in which they are embedded, they can be true or false in their own right, which implies that they encode concepts.

Th e next argument for the conceptual status of parentheticals is their po- tential compositionality: they can have a complex syntactic and semantic structure:

(39) John is, I increasingly tend to think, a fool.

(40) Th is is, I strongly suspect, despite all indications to the contrary, a Tinto- retto.

(41) I sądzę, że właśnie ta trudność sprawiła ‒ tak przynajmniej sądzę ‒ że trzeba było w tekście sprawozdania słowo zmienić.

‘And I think that this particular diffi culty – at least I think so – led to the necessary change of the word in the report content.’

(42) Warszawa kojarzyła mi się z zimą, a Sulejówek z latem, które, jak wtedy sądziłem, trwa tu także zimą.

‘I associated Warsaw with winter, whereas Sulejówek with summer, which, I thought at that time, lasted there during winter as well.’

As Ifantidou (1994, 2001) argues, the parentheticals in the utterances (39) and (40) above seem to encode concepts, which are capable of undergoing


regular compositional semantic rules. It is not clear how they can be analyzed in procedural terms.

As shown in the previous section, discourse parentheticals may contribute to the pragmatic interpretation of an utterance at diff erent levels: implicit or explicit.

Ifantidou (1994, 2001), following Wilson and Sperber (1993), argues that true parentheticals, comment clauses included, contribute to the explicit aspect of communication as well. If explicatures are taken to be explicitly communi- cated assumptions of an utterance and may include the proposition expressed by this utterance and higher-level descriptions obtained by optionally embed- d-ing this proposition under a speech-act verb or a propositional-attitude verb, then it might be proposed that parenthetical comment clauses can be analyzed as providing the hearer with explicit guidance as to the intended higher-level explicature. To obtain the higher-level explicatures the hearer must make ad- ditional inferences about the speaker’s attitude to the proposition and the type of speech-act the speaker is performing and the greater the degree of decoding involved, the more explicit the communication (Ifantidou 1994: 204).

Th us within the RT framework, the fact that parentheticals encode con- cepts, though they are non-truth conditional, can be captured since higher- -level explicatures may contribute conceptual representations (truth-condi- tional in their own right) recovered by a combination of decoding and infer- ence. For example, in (43) the parenthetical ‘I think’ is non-truth conditional and it can be taken to provide the hearer with explicit guidance as to the in- tended higher-level explicature in (44) (Ifantidou 1994: 204):

(43) John is, I think, at the airport.

(44) Mary thinks John is at the airport.

However, as Ifantidou (1994) notes, the problem with this analysis is that it assigns the same propositional structure to true parentheticals and their main-clause counterparts. It appears then that true parentheticals encode the same conceptual information as main-clauses and the claim that the addi- tion of a comment clause sentence-initial, mid-sentence and sentence-fi nal to achieve a specifi c pragmatic eff ect to weaken or strengthen the assertion is not accounted for.

One solution (Ifantidou 1994, 2001) is to treat the host and the parenthet- ical comment clause as two utterances, two separate speech-acts, one com- menting on the other, just as in the case of discourse parentheticals discussed in the fi rst section. Th is analysis might be supported by the accounts in which parenthetical expressions are taken to be phonologically, syntactically and se- mantically independent of their hosts. On this approach, (43) would assert that John is at the airport and that the speaker thinks that John is at the airport,


with the main point of the utterance being made by the most deeply embedded assertion (Ifantidou 1994: 206). Th us, intuitions about the truth conditions of the parenthetical in (43) are intuitions about the subpart that makes the major contribution to overall relevance, which constitutes its main point.

Further, the fact that parentheticals express a diminished commitment to the proposition expressed follows from the semantics of the constructions to- gether with considerations of optimal relevance provided in (45) from Ifanti- dou (1994: 206).

(45) An utterance, on a given interpretation, is optimally relevant if:

a. it achieves enough contextual eff ects to be worth the hearer’s attention;

b. it puts the hearer to no unjustifi able processing eff ort in achieving those eff ects.

Th e very fact of requesting the hearer’s attention by means of an utterance creates in him an expectation of optimal relevance. By the condition speci- fi ed in (45a), the utterance is expected to yield more eff ects than any other information the hearer could have been attending to at the time. According to the condition in (45b), the speaker who wants to eliminate any risk of being misunderstood, should make the intended interpretation as easy to recover as possible.

According to the defi nition of optimal relevance, the extra processing eff ort incurred by the parenthetical construction ‘I think’ needs to be off set by extra or diff erent contextual eff ects, which in this case, is a diminished commitment to the proposition expressed. Where the parenthetical is ‘I know,’ the extra con- textual eff ect would oft en be a strengthened commitment to the proposition expressed. If the speaker wanted to communicate a strong assertion, she would have done so and would have spared the unnecessary eff ort and avoided a mis- understanding.

6. Conclusion

In this article, I have looked at the semantic/discourse properties of paren- theticals from a relevance-theoretic perspective, which takes the pragmatic ef- fects parenthetical phenomena contribute to utterance interpretation to follow from the considerations of optimal relevance. However, if the explanations RT off ers are to be convincing, some further questions need to be answered and additional work needs to be done. For example, in a recent paper, Kaltenböck (2008) investigates the communicative uses of parenthetical comment clauses in a corpus of spoken English and establishes a link between prosodic proper- ties of parentheticals and diff erent pragmatic functions they may have. Specifi -


cally, he argues that lack of prosodic independence (i.e., comma intonation) of a parenthetical element can suggest a diff erent communicative use, i.e., a tex- tual function that can be a result of continuous semantic bleaching and repre- sent diff erent stages in an ongoing grammaticalisation (or pragmaticalization) process. As structuring devices, these comment clauses have little semantic content, they help the speaker to overcome production diffi culties and main- tain fl uency, and the hearer to signal what belongs together and to structure the information fl ow. Comment clauses that have an epistemic relationship to the proposition they modify, on the other hand, would be mostly prosodically independent of the phrase to which they are attached.6

It remains to be seen in future research if similar observations on the rela- tion between the individual prosodic patterns of comment clauses and their contribution to interpretation in discourse are universal and carry over to other languages.


Austin, J. L. (1962). How to Do Th ings with Words. Oxford: Clarendon Press.

Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan (1999). Longman Grammar of Spoken and Written English. Harlow: Pearson Educational.

Blakemore, D. (1987). Semantic Constraints on Relevance. Oxford: Blackwell.

Blakemore, D. (2006). Divisions of labour: Th e analysis of parentheticals. Lingua 116:


Blakemore, D. (2009). On the relevance of parentheticals. Actes d’IDP 09: 9‒17.

Blakemore, D. and R. Carston (2005). Th e pragmatics of sentential co-ordination with and. Lingua 115(4): 569‒89.

Bolinger, D. (1989). Intonation and its Uses. Melody in Grammar and Discourse. Lon- don: Edward Arnold.

Brinton, L. J. (2008). Th e Comment Clause in English. Syntactic Origins and Pragmatic Development. Cambridge: Cambridge University Press.

Carston, R. (2002). Th oughts and Utterances: Th e Pragmatics of Explicit Communica- tion. Oxford: Blackwell.

Dehé, N. and Y. Kavalova (2007). Parentheticals: An introduction. In N. Dehé and Y. Kavalova (eds.). Parentheticals. Amsterdam and Philadelphia: John Benjamins Publishing Company.

Dehé, N. and A. Wichmann (2010). Th e multifunctionality of epistemic parentheticals in discourse: Prosodic cues to the semantic-pragmatic boundary. Functions of Lan- guage 17 (1): 1‒28.

Espinal, T. M. (1991). Th e representation of disjunct constituents. Language 67 (4):


Grice, H. P. (1989). Studies in the Ways of Words. Cambridge: Harvard University Press.

6 A similar claim is off ered in Dehé and Wichmann (2010) and Payà (2003).


Ifantidou, E. (1994). Parentheticals and Relevance. UCL Working Papers in Linguistics 5 (1994): 193–210.

Ifantidou, E. (2001). Evidentials and Relevance. Amsterdam and Philadelphia: John Benjamins Publishing Company.

Kaltenböck, G. (2008). Prosody and function of English comment clauses. Folia Lin- guistica Acta Societatis Linguisticae Europaeae 42 (1) 2008: 83‒134.

Payà M. (2003). Prosody and pragmatics in parenthetical insertions in Catalan. Cata- lan Journal of Linguistics 2: 207‒27.

Quirk, R., S. Greenbaum, G. Leech and J. Svartvik (1985). A Comprehensive Grammar of the English Language. Harlow: Longman.

Rouchota, V. (1996). Discourse connectives: what do they link? UCL Working Papers in Linguistics 8 (1996): 1‒15.

Rouchota, V. (1998). Procedural meaning and parenthetical discourse markers. In A. Jucker and Y. Ziv (eds.). Discourse Markers: Descriptions and Th eory. Amster- dam and Philadelphia: John Benjamins Publishing Company. 97‒126.

Sperber, D. and D. Wilson (1995). Relevance: Communication and Cognition. Oxford:


Taglicht, J. (1996). Constraints on intonational phrasing in English. Journal of Linguis- tics 34 (1): 181‒211.

Urmson, J. O. (1952). Parenthetical verbs. Mind 6: 1 480‒496.

Wichmann, A. (2001). Spoken parentheticals. In K. Aijmer (ed.). A Wealth of English.

Studies in Honour of Göran Kjellmer. Göteborg: Acta Universitatis Gothoburgien- sis. 177‒93.

Wilson, D. and D. Sperber. (1993). Linguistic form and relevance. Lingua 90 (1/2):





Related subjects :